[tarek] The Perfect Running App

Stop comparing Apples and Oranges

I think this is the worst feature all running apps have: they will tell you your average pace and your "progression". Some of them try to take a coach-like tone and tell you stuff like:

You're getting slow!

Jeez. Of course I am getting slow. I ran an interval run yesterday with a specific pace of 3'40/km and today I am doing a long run at a very slow pace because this is how you train !

Yeah the pace varies during the week. Nothing I can do with this chart here.

Global metrics that use all your recorded runs have no sense

You can't do this. That does not mean anything. Knowing that my yearly average pace is 4'45 does not mean anything.

You can't either really know what will be my estimate finish on a 5k using one of my long runs.

My estimated 5k finish time based on one long run. Of course it's completely wrong.

Also, the volume of trainings and kilometers you do really depends on what you're aiming at. Someone that is getting ready for a marathon will do at least 60km/week, and will take it easy the last week.

That would be a non-sense if you are getting ready for a 5k - But in most apps, the runner that prepares for the marathon will appear like a killer compared to the 5k runner. Tell this to the runners that are doing 16' on a 5k...

I will soon lose my Gold medal since my 5 weeks training plan is over.

Anyways. My point is that the software should be smarter there. Every run needs to be classified in very specific groups to have any useful metrics on it. The most important ones are:

• long runs
• short interval runs
• long interval runs
• races

Long Runs

A long run is basically running at a lower pace for a longer time than usual. If you are getting ready for a 10k, you usually have one or two long runs per week, that will last for 1h to 1h30 tops.

The goal of those runs is to try to keep the same steady heart rate, and usually if the place where you are running is flat enough, the same pace.

A long run look like this:

The red zone is the HR drift.

There are several interesting things in this chart: you don't usually warmup when you do long runs. So the first portion of the run is a slow raise of your pace and heart rate until you've reached the targeted zone.

The quality of a long run is your ability to stick with the same heart rate for the whole session. Unless you are very careful and slow down a bit over time, there will be a slow, natural increase of your heart reate over time.

The most interesting metric in the case of a long run is to determine how flat your HR is, excluding the warmup section at the beginning.

A possible variation is to add a few strikes in the middle of your long runs. It makes it less boring. For example 2x2mn at 10k speed. It's important that these two strikes don't confuse the software that measures how flat your HR is.

Comparing long runs can be done by looking at;

• how good you are at keeping the desired HR over time
• how fast your heart is beating for a given pace as long as the circuit is flat enough and how this evolves over time.

Short Interval Run

If I do a short interval run, this is how things will go:

• 30' warmup
• 12x (45" at max speed, 30" slow)
• 10' to cool down

The red line is the linear regression of the fast strikes.

We can ditch the warmup. It does not bring any interesting data besides the volume of training. The only interesting thing to do there is to make sure it was long enough. That varies between runners, but for short intervals, it's usually roughly as long as the intervals themselves.

Now for the intervals, a quality metrics is to check if they are all done at the same speed. It's quite common to start the series very fast and to finish slowly, completely burnt by the first strikes. This is not good! A good interval run is done at the same speed for all strikes (both fast and slow segments). A great interval run is done with a slightly faster speed for the last intervals.

A good metrics in this case is the linear regression of the pace for the fast segments then the slow segments. It should be flat or slightly increasing.

The ten last minutes are also very intesting: how fast your heart rate decreases over the ten minutes ? The faster the better.

Comparing interval runs between them can be done by checking how these metrics progress over time.

Long Interval Runs

Long interval runs are like the short ones. The only difference is that you can take into account your HR variation between strikes to see how good you are at decreasing your HR between each strike. A linear regression can be added there.

Race

A Race is a very specific run, and has its specific metrics. Usually, we tend to start too fast with the danger of getting in the red zone in the first kilometers.

This is the perfect 10k run:

A 10K run with a perfect negative split.

The first 5-6k are down 3-5 seconds slower than your target pace, and the end of the run 3-5 seconds faster. This is called a negative split. The last 500m should be as fast as you can.

So for races, what I want to find out is if I was able to do a negative split, if I did not start too fast and if I was able to sprint to the finish line.

This is also a set of metrics that can be compared from race to race over time.

The Impact of Temperature

I have a friend at the racing club that trained hard for 8 weeks for a marathon. He was aiming at 3h15mn and practiced accordingly. The day the race was there, we had a very unusual heat wave in France - 37ºC which is a lot for my area. He finished the marathon in 3h40 and was happy about his performance!

The bottom line is that the heat or the cold directly impact how we perform - and this varies a lot between individuals. In my dream running app, I want to correlate my results with the temperature.

I want all my predictions to have a ponderation (not sure that's how you say it in english ;)) with the temperature.

The Impact of Rest

How long did you rest since the last run ? How did it impact your performances ?

With this information and how fast your heart slows down after your training, we can detect overtraining and undertraining.

I know Polar has a bit of this in its latest software. It tells you how long you should rest before your next run. I wonder how they calculate this.

The Social Part

Last year I ran in San Franscisco near the bay bridge with my Nike+ watch and when I uploaded my run I was delighted to see that I did a 1 mile segment many people did.

Unfortunately, the only thing the app was able to tell me is that I was 365th in terms of pace and another ridiculous rank in terms of how many runs I did there.

This is so stupid. Where am I getting with this ? Becoming the Running Mayor of the Pier? :)

There's one thing that could be interesting in running apps when comparing your data with other people: group users by ages and by records.

I am 37 and my 10k record is around 38' - I don't really care to know how I perform on a weekly basis compared to an elite runner who does 31', or a more casual runner who does 50'.

What could be interesting though is to compare with people that are at my level or age and that are getting ready for the same race maybe ? or a race that's similar enough and close enough.

Conclusion

This blog post is just a brain dump - some ideas are pretty vague and I have not really talked about the UX of the Running Software.

But the bottom line is that if you don't just jog, and want to use a running application for serious training, what I have described is what I think is needed.

I hope Smashrun will go into that direction!